AI brokers would pay for computing energy utilizing the token, due to this fact straight linking its demand to community utilization.
Arthur Hayes has expanded his thesis for Flop Labs, an AI cost mission he introduced lately, arguing that AI brokers want a type of cash that may be exchanged straight for computing energy somewhat than counting on {dollars}, Bitcoin, or standard cost rails.
The thought is straightforward on paper however formidable in follow: if AI brokers grow to be main customers of computing energy, he believes their cash must be straight redeemable for the sources they really use.
A Case for a Compute-Based mostly Foreign money
Flop Labs laid out Hayes’ newest argument in a six-part thread on August 27, beginning with a primary drawback: there is no such thing as a environment friendly spot marketplace for turning cash right into a identified amount of compute over a identified interval.
That’s essential as a result of AI brokers have totally different spending wants than folks. “Brokers don’t eat. They eat floating-point operations,” Flop Labs wrote, referring to the calculations required to run AI fashions.
The argument follows that an agent’s cash must be helpful for the factor the agent truly consumes. Hayes’ view, as introduced by Flop Labs, is that the foreign money able to changing into compute on demand at a good value might grow to be the cash utilized by an agent financial system.
He additionally questioned whether or not stablecoins and tokenized playing cards are suited to that function, provided that these techniques are designed round establishments and customers which have folks, authorized entities, and physical-world wants behind them. An autonomous agent has none of these issues.
The proposed Flop Community is designed round that distinction. GPU operators would offer inference and obtain FLOP, whereas validators would test the work cryptographically. Miners would additionally publish a stake that could possibly be misplaced in the event that they submit false outcomes. Brokers would then pay for compute utilizing the identical token they maintain, with the community offering proof that the requested work was delivered.
You may additionally like:
Per the mission’s tokenomics, that are nonetheless preliminary, the FLOP provide ought to hit about 17.2 billion by 12 months 10 of its existence, with no enterprise capital allocation or presale. The Genesis airdrop is listed at 3.5 billion tokens, together with 1.5 billion for miners, 1.2 billion for brokers, 310 million for validators, and 790 million for reserves and incentives. There’s additionally a deliberate testnet within the works, which is predicted to run for about 90 days, with the supply code public.
Connecting AI Debt to a Crypto Liquidity Guess
The opposite a part of the thread is extra acquainted to anybody who has adopted Hayes’ AI criticism. He has spent months calling AI funding a bubble, however he mentioned the surplus sits in information middle debt and unprofitable hyperscaler shares, not in agentic know-how itself.
For that cause, the BitMEX co-founder expects AI spending to decelerate subsequent 12 months, then contract, forcing bailouts greater than these seen in 2008, which he believes will push new cash towards crypto, probably sending Bitcoin towards $1 million.
Nonetheless, real-world utilization is lagging the pitch, with analyst Jamie Coutts lately finding that settlement quantity on Coinbase’s x402 agent cost protocol had gone down 93% this 12 months. Whereas he known as it a “actuality test” for these constructing within the area, he expects quantity to as soon as once more choose up within the fourth quarter.
